Controversies about sugars: results from systematic reviews and meta-analyses on obesity, cardiometabolic disease and diabetes

Abstract

Fructose-containing sugars are a focus of attention as a public health target for their putative role in obesity and cardiometabolic disease including diabetes. The fructose moiety is singled out to be the primary driver for the harms of sugars due to its unique endocrine signal and pathophysiological role. However, this is only supported by ecological studies, animal models of overfeeding and select human intervention studies with supraphysiological doses or lack of control for energy. The highest level of evidence from systematic reviews and meta-analyses of controlled trials has not shown that fructose-containing sugars behave any differently from other forms of digestible carbohydrates. Fructose-containing sugars can only lead to weight gain and other unintended harms on cardiometabolic risk factors insofar as the excess calories they provide. Prospective cohort studies, which provide the strongest observational evidence, have shown an association between fructose-containing sugars and cardiometabolic risk including weight gain, cardiovascular disease outcomes and diabetes only when restricted to sugar-sweetened beverages and not for sugars from other sources. In fact, sugar-sweetened beverages are a marker of an unhealthy lifestyle and their drinkers consume more calories, exercise less, smoke more and have a poor dietary pattern. The potential for overconsumption of sugars in the form of sugary foods and drinks makes targeting sugars, as a source of excess calories, a prudent strategy. However, sugar content should not be the sole determinant of a healthy diet. There are many other factors in the diet-some providing excess calories while others provide beneficial nutrients. Rather than just focusing on one energy source, we should consider the whole diet for health benefits.

Keywords: Cardiovascular disease; Diabetes; Fructose; Obesity; Overweight; Review; Sugars.

Fig. 1

Hierarchy of evidence in evidence-based medicine

Fig. 2

Sources of sugars and incident type 2 diabetes. Adapted from [68]. Summary estimates (diamonds) were derived from pooled risk ratios for comparison of extreme quantiles (the highest level of exposure compared with the lowest level of exposure). The one exception was for cakes and cookies, which compared the highest level of exposure with the middle level of exposure; the reference exposure that was associated with the lowest risk. Data are expressed as risk ratios with 95% CIs. Asterisks indicate significant interstudy heterogeneity as assessed by the Cochran Q statistic and quantified by the I ² statistic at a significance level of P < .10. SSBs sugar-sweetened beverages

Fig. 3

Bodyweight changes (kg) over a 4-year period associated with an increase in the consumption of different food items. Using data from the Nurses’ Health Study, the Nurses’ Health Study II and the Health Professionals Follow-up Study as reported by Mozaffarian et al. [105]. Increased consumption is based on servings/day for all items except trans fat (per cent total energy) and fried foods (servings/week). Data represent pooled mean changes with 95% confidence intervals adjusted for age, baseline body mass index at the start of each 4-year interval, sleep duration and changes in physical activity, smoking, alcohol use, television watching and each additional food item

Fig. 4

Substitution trials. The meta-analyses are of isocaloric substitution trials, in which fructose was exchanged for other carbohydrate sources under energy-matched conditions. Summary estimates (diamonds) were derived from pooled trial-level data. To allow the summary estimates for each endpoint to be displayed on the same axis, mean differences were transformed to standardized mean differences (SMDs). Pseudo-95% CIs for each transformed SMD were derived directly from the original mean difference and 95% CI. The scales were also flipped for high-density lipoprotein cholesterol (HDL-C), whole-body insulin sensitivity and hepatic insulin sensitivity so that the direction of the effect for benefit or harm was in the same direction as that for the other endpoints. Asterisks indicate significant interstudy heterogeneity as assessed by the Cochran Q statistic and quantified by the I ² statistic at a significance level of P < .10 (the higher significance level was chosen owing to the poor sensitivity of the test). ALT alanine aminotransferase, Apo-B apolipoprotein B, DBP diastolic blood pressure, FBG fasting blood glucose, FBI fasting blood insulin, GBP glycated blood proteins, HOMA-IR homoeostatic model assessment-insulin resistance, IHCL intrahepatocellular lipid, LDL-C low-density lipoprotein cholesterol, MAP mean arterial pressure, SBP systolic blood pressure, TG triglycerides, No. total number of participants included in the meta-analysis of the controlled dietary trials

Fig. 5

Addition trials. The meta-analyses are of hypercaloric addition trials, in which excess calories from fructose were added to a diet compared with the same diet without the excess calories. Summary estimates (diamonds) were derived from pooled trial-level data. To allow the summary estimates for each endpoint to be displayed on the same axis, mean differences were transformed to standardized mean differences (SMDs). Pseudo-95% CIs for each transformed SMD were derived directly from the original mean difference and 95% CI. The scales were also flipped for high-density lipoprotein cholesterol (HDL-C), whole-body insulin sensitivity and hepatic insulin sensitivity so that the direction of the effect for benefit or harm was in the same direction as that for the other endpoints. Asterisks indicate significant interstudy heterogeneity as assessed by the Cochran Q statistic and quantified by the I ² statistic at a significance level of P < .10 (the higher significance level was chosen owing to the poor sensitivity of the test). ALT alanine aminotransferase, Apo-B apolipoprotein B, DBP diastolic blood pressure, FBG fasting blood glucose, FBI fasting blood insulin, GBP glycated blood proteins, HOMA-IR homoeostatic model assessment-insulin resistance, IHCL intrahepatocellular lipid, LDL-C low-density lipoprotein cholesterol, MAP mean arterial pressure, SBP systolic blood pressure, TG triglycerides, No. total number of participants included in the meta-analysis of the controlled dietary trials

Fig. 6

Fructose-containing sugars and weight change in controlled dietary trials. Forest plots of summary estimates from recent meta-analyses of the effect of different fructose-containing sugars interventions on indices of body weight in controlled dietary trials involving children and adults. The meta-analyses were grouped broadly based on the interventions in question: isocaloric sugar substitution interventions, in which sugars were exchanged for other carbohydrate sources under energy-matched conditions; sugar supplementation interventions, in which sugars supplement background diets providing excess energy compared with the background diets alone without the excess energy; and sugar reduction interventions, in which excess from sugars is reduced in background diets compared with the background diets still containing the sugars. Indices of body weight included body weight in Sievenpiper et al. [107], Kaiser et al. [140] and Te Morenga et al. [49]. For isocaloric sugar substitution only; body fatness in Te Morenga et al. [49] for all other comparisons; and BMI z scores in Malik et al. [55]. Summary estimates (diamonds) were derived from pooled trial-level data. To allow the summary estimates for each endpoint to be displayed on the same axis, mean differences (MDs) were transformed to standardized means differences (SMDs). Pseudo-95% confidence intervals (CI) for each transformed SMD were derived directly from the original MD and 95% CI. Asterisks indicate significant interstudy heterogeneity as assessed by the Cochran Q statistic and quantified by the I ²-statistic at a significance level of P < .10. SSBs sugar-sweetened beverages, No. total number of participants included in the meta-analysis of the controlled dietary trials